[time-nuts] Ground loops in measurements?

David davidwhess at gmail.com
Mon May 16 06:18:21 UTC 2016


I will try to avoid too much repetition of what has already been
posted.  Some other ways I have handled ground loops:

1. Capacitive or inductive coupling - this may include AC coupling in
*both* the center conductor and shield of a coaxial cable.

2. High isolation power supplies - these may use special transformers
which include a large air gap to reduce capacitive coupling.  In
extreme applications, they can also be implemented using ceramic
transducers.  A less expensive way is to implement a small high
frequency inverter using a torroidal transformer with well separated
windings; Tektronix liked to do this and I have a number of older DMMs
that use this for isolation of their floating inputs.

3. Some instruments implement isolation internally and use connectors
which are isolated from chassis ground.  The Tektronix PG506 pulse
generator is a good example of this.  10Base-2 used isolated BNC
connectors and an isolated power supply.

4. USB is a major and sometimes destructive problem with its shield
and signal ground connection between instruments.  USB isolated hubs
are available and Analog Devices even makes an ASIC for this now.
Ethernet is pretty good in this respect.  One reason I have always
liked RS-232 is that it was relatively easy to isolate.  The old MIDI
standard uses optical isolation as part of the standard.

5. One trick you can find inside test instrumentation is the use of
common mode chokes (current baluns) on transmission lines between
modules.  The same thing can be done externally.

6. Instrumentation amplifiers can be used in place of transformers for
isolation in some cases.  Differential inputs can be very handy to
prevent ground loops even from single ended sources.

7. One trick I have used at lower frequencies is to use currents
instead of voltages (transconductance outputs) with resistive
terminations at the endpoint and the coax only grounded at one end.
This also works across a ground plain when ground currents would
otherwise corrupt precision like maybe if you were driving the control
voltage input of an OCXO.

8. Differential oscilloscopes probes are very handy for tracking down
ground loops.  I really like my Tektronix 7A13 and 7A22 differential
amplifiers for this.

9. If you are making a normal single ended measurement with an
oscilloscope probe, measure the point where the ground clip is
attached; if the signal is not flat, then you have a ground loop.



More information about the Time-nuts_lists.febo.com mailing list